Last active
March 27, 2019 11:00
-
-
Save AidanRocke/33c1d5268d8f8c3b395cc81ba6397f47 to your computer and use it in GitHub Desktop.
A generative model based on the 2/3 power law
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
using Random | |
function crazy_paths(N,delta_t) | |
ddx, dx = 10*(2*rand(N+1) .-1.0), 10*(2*rand(N+1) .-1.0) | |
ddy, dy = 10*(2*rand(N+1) .-1.0), 10*(2*rand(N+1) .-1.0) | |
x, y = 10*(2*rand(N+1) .-1.0), 10*(2*rand(N+1) .-1.0) | |
K = abs(ddx[1]*dy[1]-dx[1]*ddy[1]) | |
A, B = zeros(N), zeros(N) | |
for i = 1:N | |
## the transformation: | |
alpha, beta, theta = (-1)^rand(1:2)*rand(0.1:0.01:10), (-1)^rand(1:2)*rand(0.1:0.01:10), rand() | |
delta = 2*rand(2) .-1.0 ## a translation vector which doesn't change the volume | |
M = [cos(2*pi*theta)/alpha sin(2*pi*theta)*beta; -1*sin(2*pi*theta)/beta cos(2*pi*theta)*alpha]; | |
A[i], B[i] = M[1], M[3] | |
## update the variables: | |
#ddx[i+1] = M[1]*ddx[i] + M[3]*dx[i] .+ delta[1] | |
#ddy[i+1] = M[1]*ddy[i] + M[3]*dy[i] .+ delta[1] | |
#dx[i+1] = M[2]*ddx[i] + M[4]*dx[i] .+ delta[2] | |
#dy[i+1] = M[2]*ddy[i] + M[4]*dy[i] .+ delta[2] | |
## update the variables: | |
ddx[i+1] = M[1]*ddx[i] + M[3]*dx[i] | |
ddy[i+1] = M[1]*ddy[i] + M[3]*dy[i] | |
dx[i+1] = M[2]*ddx[i] + M[4]*dx[i] | |
dy[i+1] = M[2]*ddy[i] + M[4]*dy[i] | |
## update: | |
x[i+1] = x[i] + dx[i]*delta_t + 0.5*ddx[i]*delta_t^2 | |
y[i+1] = y[i] + dy[i]*delta_t + 0.5*ddy[i]*delta_t^2 | |
end | |
q_4 = ddy[1]*A[1] + dy[1]*B[1] | |
q_2 = ddx[1]*A[1] + dx[1]*B[1] | |
return x,y,q_4, q_2 | |
end | |
function test_hypothesis(N,delta_t) | |
output = zeros(100) | |
for i=1:100 | |
x,y,q_4, q_2 = crazy_paths(N,delta_t) | |
range = rand(20:80) | |
lower, upper = rand(1:N-range), rand(N-range+1:N) | |
expected_slope = q_4/q_2 | |
actual_slope = (y[upper]-y[lower])/(x[upper]-x[lower]) | |
max_slope = max(expected_slope,actual_slope) | |
min_slope = min(expected_slope,actual_slope) | |
sign_exp, sign_act = abs(expected_slope)/expected_slope, abs(actual_slope)/actual_slope | |
if sign_exp*sign_act == 1.0 && max_slope/min_slope < 2.0 | |
output[i] = 1.0 | |
end | |
end | |
return output | |
end | |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment